OnAuto
Auto is a playbook automation platform that lets business and marketing teams turn their operational methods into reusable, executable workflows using their existing AI agents. The platform provides a persistent "brain" of functions—reasoning, computation, prediction, recall, and interaction—coordinated into multi-step workflows with human approval checkpoints. Auto stores playbook definitions and run history outside the conversation, enabling scheduled, auditable execution of business processes.
- Enterprise Software
- Developer Tools
- Artificial Intelligence
- AI Agents
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
AI-driven development tools often make decisions that aren't visible to the user, leading to a "build drift" where the final product doesn't match the intended workflows, rules, or outcomes. Prototypes and prompts are incomplete sources of truth, and without a clear record of decisions, teams struggle to recover the product's intended behavior and ship something that meets real user needs. Similarly, recurring business operations executed manually or through ad-hoc agent prompts lack a persistent, auditable method, making it hard to ensure consistency, compliance, and quality.
Solution
Auto provides a platform that makes the workflows, rules, and product decisions behind an application visible and reviewable. It creates a persistent product model that captures capabilities, goals, outcomes, and behaviors, allowing users to inspect, correct, and approve what Auto understood before any building begins. The platform also supports "playbooks"—reusable, executable ways of working around defined business responsibilities. A playbook contains a brain (a collection of reusable functions) and workflows that coordinate those functions into steps. Users define these methods through their existing AI agents, and Auto stores the definitions and run history outside the conversation, enabling reliable, repeatable execution. Proposed changes are never applied without explicit human approval.
Target Audience
Primary customers are business and marketing specialists who know how a responsibility should be carried out, alongside engineers who support integration, reliability, and production operation. The platform targets teams that need to move from prototype or ad-hoc processes to a production-ready, auditable workflow.
Features
- Product model and playbooks that make capabilities, goals, outcomes, and behaviors visible and persistent
- Review proposed changes in context, with the ability to accept or reject before applying
- Reusable function types: Reasoning, Interaction, Prediction, Recall, and Computation
- Workflow coordination that sequences functions into multi-step processes with human handoffs
- MCP-compatible integrations for agents like Claude Code, Claude Desktop, and Codex
- Self-hosted runtime with source-available licensing (Elastic License 2.0)
- Separate display of proposed changes from the currently approved product state
- Scheduled execution of playbooks independent of the conversation, with results delivered to connected services like Slack